Licensing Camera-Sensed Audience Data for Retail Media 2026

The Evolution of Retail Media Audience Licensing in 2026

Physical retail environments are undergoing a massive transformation into data-rich advertising networks. For decades, brick-and-mortar stores relied on endcap displays, static printed signage, and traditional point-of-purchase materials that offered brands little to no measurable return on ad spend (ROAS). Today, however, the convergence of high-definition digital displays, edge computing, and advanced sensor technology has fundamentally altered the landscape. Retailers are no longer just merchants; they are becoming powerful media publishers in their own right, capable of delivering highly targeted, contextually relevant advertisements directly at the point of decision.

As highlighted by Grand View Research, the rapid expansion of Retail Media Networks (RMNs) has driven a surge in demand for granular, real-time in-store analytics. This exponential growth is fueled by the deprecation of third-party cookies and the increasing difficulty of tracking consumers across the digital web, which has pushed advertising budgets toward environments where first-party data and physical presence intersect.

Brands are no longer satisfied with static footfall estimates or historical traffic models. In the past, media buyers were forced to rely on monthly averages, seasonal traffic assumptions, or delayed reports that provided only a vague estimation of who might have been in the store. They require deterministic, real-time data to justify their programmatic ad spend. Advertisers now demand the same level of accountability, transparency, and optimization capabilities in physical stores that they have come to expect from digital channels like search and social media.

To formalize this rapidly maturing space, the Interactive Advertising Bureau (IAB) and the Media Rating Council (MRC) established unified Retail Media Measurement Guidelines. These comprehensive frameworks were designed to eliminate the wild-west nature of early in-store analytics, providing a common language and rigorous methodology for buyers and sellers alike. These critical standards define key retail media network audience metrics for the modern era.

They introduce standardized definitions like Opportunity to See (OTS) and Likelihood to See (LTS) to replace outdated traffic models with verifiable exposure data. OTS ensures that an individual was actually present in the physical zone where the screen is visible, while LTS goes a step further by utilizing sensor data to calculate the mathematical probability that the consumer's eyes actually registered the ad content.

For DOOH audience data buyers, this fundamental shift makes retail media audience licensing an absolute necessity. It is no longer an optional add-on; it is the foundational requirement for participating in the modern retail media ecosystem. It is the core mechanism for executing programmatic bidding and closed-loop attribution in physical spaces. By licensing this data, buyers can finally connect an in-store ad exposure directly to a subsequent point-of-sale transaction. Without licensed, high-fidelity audience data, retail media networks simply cannot compete with the precision of online advertising.

---

The Mechanics of Camera-Sensed Audience Data

Industry platforms leverage camera-sensed audience data via on-device computer vision to capture real-time demographics, dwell time, and attention metrics at the edge. By utilizing powerful Neural Processing Units (NPUs) embedded directly within the digital signage hardware, the heavy lifting of visual analysis is performed locally, in milliseconds, without the latency or privacy risks associated with streaming live video feeds to external servers.

Trillboards operates as a premier screen aggregator, screen network and audience measurement platform, providing the foundational data layer for this ecosystem. The platform is designed from the ground up to synthesize complex physical interactions into structured, actionable data streams that demand-side platforms (DSPs) can instantly interpret. The data collection process is rigorous, continuous, and highly detailed.

Screens measure the people and devices near them with extreme precision. As a consumer enters the field of view, the edge-computing module instantly initiates a localized tracking session. A screen keeps a record for each person its camera detects: estimated age range, gender, dominant emotion, dwell time and time spent looking at the screen. This means the system can differentiate between a shopper who quickly glances at a display while walking briskly past, and a highly engaged consumer who stops, smiles, and watches a 15-second video advertisement in its entirety. These initial on-device logs capture the exact attention metrics required by modern advertisers.

On the full sensing profile it also computes a face-identity template (a vector of 512 numbers) that Trillboards uses to recognise the same person across screens and days. This process does not store a photograph of the person; rather, it distills the unique spatial relationships of facial features into a one-way cryptographic hash. This 512-number vector acts as a mathematical representation of facial geometry, allowing for deterministic cross-screen tracking. If a shopper views an ad on a digital endcap in aisle three, and then encounters another screen at the checkout register twenty minutes later, the system recognizes the vector match and can apply intelligent frequency capping or sequential messaging.

Furthermore, Trillboards links each person to the device keys of the phones seen with them and to mobile advertising IDs. By utilizing temporal-spatial matching algorithms, the platform correlates the visual presence of the consumer with the simultaneous radio-frequency emissions of their smartphone. This vital linkage directly connects physical presence to digital identity, bridging the gap between in-store behavior and mobile targeting. Advertisers can now retarget a consumer on their mobile device an hour after they were observed physically dwelling in front of a specific product display.

---

Cloud Processing and Google Vertex AI Integration

While edge processing handles immediate attention metrics, deep contextual analysis occurs in the cloud. The edge devices are optimized for speed and privacy, filtering out the noise and transmitting only the most valuable, anonymized data points. Selected camera frames go to a cloud model (Google Vertex AI) that estimates ethnicity, age band, gender, social composition, purchase-intent stage, activity, attire, carried items, lifestyle segments and mood.

This advanced cloud processing extracts a wealth of contextual and demographic data that powers highly targeted programmatic campaigns. By leveraging the massive computational power of Google Vertex AI, the platform can understand complex human behaviors and environmental contexts that simple edge models might miss. For example, detecting "carried items" allows a retailer to trigger complementary product ads on nearby screens. If the AI detects a shopper carrying a bag of charcoal, the nearest screen can instantly bid on and display an advertisement for barbecue sauce, grilling tools, or premium meats, capturing the consumer at the exact moment of complementary need.

Analyzing "purchase-intent stage" enables brands to bid higher for consumers actively examining products versus those simply passing by. A consumer standing still, physically interacting with a product on a shelf, represents a significantly higher value to an advertiser than someone navigating the main aisle with a shopping cart. Trillboards keeps those frames for 400 days, and the frames captured between 5 May and 21 June 2026 are kept as a model-evaluation set with no scheduled deletion.

This massive data retention allows for continuous machine learning model refinement and historical audience auditing. Machine learning models are susceptible to drift over time due to changing fashions, seasonal attire variations, and evolving store layouts. By maintaining this rigorous holdout set, data scientists can continuously benchmark the AI's accuracy, ensuring that the demographic and contextual estimates remain pristine year over year.

In addition to visual data, auditory context is also measured. The ambient environment provides critical clues about consumer needs and store conditions. Speech is transcribed on the device and Trillboards stores the transcript. This allows the system to detect localized keywords,such as a customer asking a store associate for "allergy medication",which can anonymously trigger relevant programmatic bids for pharmaceutical brands. No video or audio recording is stored or transmitted, ensuring bandwidth efficiency while maintaining data fidelity. This strict adherence to text-only transcription ensures absolute compliance with global wiretapping and privacy regulations while still unlocking the immense value of auditory context.

---

Radio Sensing, GPS, and Data Partnerships

Visual data is only one half of the measurement equation for DOOH audience data buyers. To build a truly comprehensive understanding of the retail environment, visual data must be fused with robust spatial and radio-frequency intelligence. Screens send Trillboards the Bluetooth, Wi-Fi and local-network identifiers nearby phones broadcast, and their GPS position; Trillboards turns radio and GPS sensing on by default.

This creates a dense, overlapping mesh of physical and digital identifiers in the retail environment. Every smartphone constantly emits probe requests searching for known Wi-Fi networks, alongside Bluetooth Low Energy (BLE) beacons. By capturing these signals, the screens act as localized sensor hubs, mapping the precise physical coordinates of digital devices within the store's footprint. Bid requests carry the screen's IP address, location and live audience measurements.

These enriched bid requests are received by DOOH exchanges and demand-side platforms. Instead of receiving a generic request to display an ad on "Screen 4 in Store 12," the DSP receives a hyper-enriched payload detailing that there are currently three adults present, one of whom has a matched mobile identifier indicating a high propensity for luxury goods, standing precisely 1.5 meters from the display. This enables highly targeted programmatic buying based on real-time physical presence and verified attention.

Trillboards sells and shares audience measurement and location data with its data partners, including 375ai and Intuizi. These strategic alliances amplify the utility of the raw data, transforming it into macroscopic consumer insights. These partners license the data to enrich their own identity graphs and spatial analytics platforms. By ingesting Trillboards' deterministic sensor data, these platforms can refine their omnichannel attribution models, providing brands with a unified view of the customer journey from the living room television to the retail store aisle.

The data provided is highly granular and deterministic, allowing data buyers to map complex consumer journeys across the physical world. Trillboards explicitly identifies, recognizes, tracks, and links people across retail environments to provide maximum value to data buyers. This level of deterministic linkage is what ultimately allows a brand to prove that an in-store digital out-of-home campaign directly caused a measurable lift in mobile app downloads or subsequent online purchases.

---

Network Scale and the Screen Sensing Census

The value of retail media audience licensing is directly proportional to the scale and density of the measurement network. A highly intelligent screen is of limited use to a global media buyer if it only exists in a single isolated location. Advertisers require massive, standardized inventory to execute national and international campaigns efficiently. Trillboards offers unprecedented reach for programmatic buyers and data licensees.

There are 757,000+ screens reachable through Trillboards and 29,000+ screens registered on our own network. This immense footprint spans 39 countries and 2,944 distinct cities. The inventory encompasses 38 venue categories under the IAB OpenOOH taxonomy. This includes high-dwell-time environments like gyms, medical waiting rooms, and transit hubs, as well as high-intent environments like grocery stores, pharmacies, and automotive service centers. This diversity allows buyers to target consumers across the entire path to purchase, from transit hubs to point-of-sale registers.

A critical component for DOOH audience data buyers is the screen sensing census. This census represents the absolute gold standard of verified, sensor-backed inventory within the broader network. The network includes 4,000+ screens that measured their audience by camera or nearby-phone radio in a single day. This guarantees that buyers are bidding on screens that are actively alive, functioning, and actively validating human presence, completely eliminating the risk of bidding on screens that are turned off, broken, or facing an empty room.

Buyers can specifically target these high-fidelity signals as segtax=600 audience segments in their programmatic campaigns. By utilizing this specific segment taxonomy within the OpenRTB protocol, DSPs can automatically route their highest-value bids exclusively to screens that provide deterministic proof of play and proof of audience.

The depth of audience categorization is equally robust. The platform features 1,558 IAB Audience Taxonomy 1.1 nodes (segtax=4) supported in OpenRTB requests. This allows buyers to map their existing digital audience strategies directly onto the physical world without having to learn a proprietary categorization system. Furthermore, buyers benefit from 191 IAB audience segments observed in live impressions in a trailing 30-day window. This ensures that buyers are targeting actively verified consumer segments, not stale historical models. If an audience segment hasn't been physically observed interacting with the network within the last month, it is pruned from the active targeting pool, ensuring maximum ROAS for the advertiser.

---

Programmatic Infrastructure and OpenRTB 2.6

Trillboards provides the programmatic infrastructure required to activate camera-sensed audience data at scale. Collecting the data is only the first step; the true technological achievement lies in translating that data into standard programmatic protocols that the global advertising ecosystem can bid on in real-time. The platform features full OpenRTB 2.6 compatibility for both screen inventory and audience signals. This latest iteration of the OpenRTB protocol introduces advanced features like structured pod bidding, which allows digital signage operators to sell commercial breaks in a manner identical to Connected TV (CTV), maximizing yield for multi-ad slots.

SSPs and exchanges are complementary partners that connect screen inventory to programmatic buyers. Trillboards does not seek to disintermediate these vital platforms; rather, Trillboards acts exclusively as the screen aggregator and audience measurement layer, empowering these partners with enriched supply. By injecting deterministic audience data into the bid stream, Trillboards elevates the value of the SSP's inventory, driving higher CPMs and better outcomes for all stakeholders.

The platform integrates deeply with Google Ad Manager (GAM). This allows enterprise publishers who already utilize GAM for their web and mobile properties to seamlessly extend their ad operations into the physical realm. This integration utilizes IVT-compliant VAST tags and the OpenOOH venue taxonomy for seamless campaign execution. To ensure absolute measurement integrity, Trillboards implements the OM SDK (Open Measurement).

This provides MRC-compliant ad verification for every impression served, guaranteeing viewability standards are met. Brands can independently verify that their video creative actually rendered on the screen, played to completion, and was not obscured or muted. Supply chain transparency is a core operational mandate in an era where advertisers are increasingly scrutinizing the path their money takes.

The platform executes a rigorous 14-check OpenRTB 2.6 supply-chain validation runbook. This includes real-time validation of sellers.json, ads.txt, and schain in every single VAST request. This exhaustive validation process guarantees that buyers are purchasing authorized inventory, completely eliminating the risk of domain spoofing, unauthorized reselling, and digital out-of-home fraud. To maintain network integrity, Trillboards has executed 255,234 creative-level classifications performed across 45 IAB Content Taxonomy top-level categories. This ensures that family-friendly retail environments are never exposed to inappropriate ad creative, providing brand safety guarantees that rival the strictest premium publisher environments.

---

API-First Architecture for AI Agents and Developers

For AI agents and the developers who build buying agents, Trillboards offers a truly API-first architecture. The modern ad tech stack is increasingly automated, moving away from manual media planning toward algorithmic execution. The Partner SDK (@trillboards/ads-sdk) serves as programmatic infrastructure-as-a-service. It abstracts away the immense complexity of hardware integration, sensor fusion, and programmatic bidding into a clean, modern developer experience.

It supports TypeScript, React, React Native, Flutter, CTV, and server-side implementations. Digital signage software companies can easily add ad monetization via this SDK without building complex ad tech from scratch. Instead of spending millions of dollars and years of engineering time attempting to build an OpenRTB-compliant ad server, developers can drop the SDK into their existing codebase and begin monetizing their screens in a matter of days. CTV and smart display OEMs use it to build out native ad infrastructure directly into their hardware.

The complete OpenAPI reference is available at https://api.trillboards.com/docs. This full REST API governs device, audience, venue, webhook, and analytics management. It allows enterprise clients to programmatically provision thousands of screens, update venue metadata, and pull granular performance reports without ever logging into a graphical user interface. Trillboards also hosts an MCP server at https://api.trillboards.com/mcp.

This functions as an AdCP sales agent that an AI agent can plan and buy through autonomously. As large language models (LLMs) and autonomous AI agents become the new standard for media buying, this endpoint allows an AI to natively query inventory, negotiate pricing, and execute campaigns using natural language or structured JSON commands. Data delivery is managed through a powerful webhook-driven event architecture. Subscribers receive real-time push notifications for device status changes, impressions, payouts, and audience spikes.

API access is structured across three tiers to support varying enterprise needs. The Basic tier provides 200 requests per minute, ideal for small network operators testing the waters of programmatic monetization. The Developer tier unlocks 1,000 requests per minute and full venue intelligence, catering to mid-sized networks requiring deeper analytics and faster data synchronization. The Enterprise tier offers 5,000 requests per minute, raw data exports, and a dedicated SLA for high-volume data buyers, ensuring that massive identity graph providers and global DSPs have the uninterrupted firehose of data they require to power their platforms.

---

Publisher Economics: The 60/40 Revenue Split

For media enterprises, retail media networks, and large publisher networks managing 100+ screens, the economics are highly favorable. Historically, digital signage operators were burdened by expensive, per-screen software licensing fees that severely ate into their profit margins before a single ad was even sold. Trillboards provides free screen-management and ad-serving software. Publishers pay exactly $0/screen/month for the enterprise-grade platform.

Free for publishers, the platform monetizes through ad demand, not publisher fees. This radical shift in the business model eliminates the upfront capital expenditure of deploying a retail media network, allowing retailers to scale their screen footprint aggressively. When programmatic inventory is sold, the revenue share is strictly defined. Programmatic ad revenue is split 60/40 in the publisher's favor: the venue/publisher keeps 60%, Trillboards keeps 40%.

This transparent model aligns the platform's success directly with the publisher's yield. Trillboards only makes money when the publisher makes money, incentivizing the platform to constantly optimize routing, improve audience measurement accuracy, and attract the highest-paying premium demand. Retailers and property management companies can instantly transform passive screens into active revenue centers.

Innovative networks are already capitalizing on this model by deploying smart screens in high-traffic retail environments, as VapeTM demonstrates with their smart vending machines with digital displays and advertising screens. By integrating audience sensing directly into the point of sale, these machines generate dual revenue streams: product sales and highly targeted programmatic advertising. Similarly, experiential retail concepts are integrating digital signage, as Sweet Robo demonstrates with their robotic vending machines (cotton candy, ice cream, popcorn) featuring interactive screens. These high-engagement environments naturally capture extended dwell times, making them incredibly lucrative for video advertisers.

Pharmacy owners can learn more in our /guides/pharmacy-digital-signage-income/ documentation, or explore other venues in our /guides/ hub. These specialized guides provide step-by-step blueprints for maximizing yield in specific retail verticals, acknowledging that a pharmacy requires a vastly different content and advertising strategy than a robotic cotton candy machine. These deployments benefit directly from robust retail media network audience metrics. By utilizing Trillboards' screen onboarding, playback, and venue controls, operators maximize their fill rates. SSPs and exchanges then connect this premium, data-enriched inventory to global programmatic demand, ensuring that every available ad slot is contested by multiple high-value bidders.

---

The Future of DOOH Audience Data Buyers

The landscape for DOOH audience data buyers is evolving rapidly. The days of buying bulk impressions based on static, unverified location data are effectively over. The modern media buyer demands absolute certainty, requiring platforms that can mathematically prove the presence and attention of their target demographic. Industry publications like Ariadne note that AI-enabled cameras offer a highly detailed screen sensing census for demographic estimation.

However, the transition requires sophisticated data infrastructure to manage effectively. It is not enough to simply mount a camera to a screen; the resulting data must be processed, anonymized, standardized, and injected into the programmatic bid stream in real-time. Buyers can view published floor CPMs for every audience segment directly at /pricing/. This unprecedented transparency allows agencies to accurately forecast campaign costs and build precise media plans without engaging in lengthy, opaque negotiation cycles.

Campaign planning is streamlined through the interactive planner at /plan/. Here, buyers can visually map out their target demographics, filter by venue types, and instantly see the available scale and estimated reach of their desired audience across the global network. Advertisers can sign up and access deal IDs for their own DSPs at /ads/auth. This self-serve infrastructure empowers trading desks to activate retail media campaigns within minutes, seamlessly integrating physical screens into their existing omnichannel workflows.

By licensing verified, audited audience data, modern brands transition from estimated reach to guaranteed attention. The combination of Google Vertex AI processing, 512-vector facial templates, and deterministic MAID linkage provides unmatched clarity. Advertisers no longer have to guess if their message resonated; they have the cryptographic and sensor-backed proof that their target consumer stood in front of the screen, looked at the creative, and subsequently engaged with the brand on their mobile device.

Trillboards ensures that every bid request is enriched with live audience intelligence. From demographics and dwell time to attention metrics and foot traffic, the data is comprehensive. This level of real-time audience intelligence redefines retail media strategies for 2026 and beyond. Media buyers, data measurement teams, and AI bidding agents finally have the raw data required to optimize physical space advertising, bringing the precision, accountability, and performance of digital marketing into the real world.